Running Workflows
Execute your pipeline on one image or a full batch. Monitor each step, pause when needed, and retry failures.
Single image execution
Upload or select an image and click Run. The workflow processes each step in order, and you see the result update on the canvas after every step. This is the fastest way to test a new pipeline or process a one-off image.
The canvas highlights the currently active step so you always know where your image is in the pipeline.
Batch execution
Switch to batch mode to run the same workflow on many images at once. Add images the same way as regular batch processing: upload files, paste URLs, or import from a Google Sheet.
Each image runs through the full pipeline independently. If one image fails at step 3, the others keep going. A progress dashboard shows the overall batch status plus per-image details.
Pause and resume
Click Pause to stop the workflow after the current step finishes. No work is lost. Images that were mid-step will complete that step, then wait.
Hit Resume to pick up exactly where you left off. Pausing is useful if you want to spot-check results partway through a large batch before committing to the rest.
Retrying failed steps
Failed steps show a red indicator on the canvas with the error message. Common causes: a source image URL went offline, the AI model timed out, or a file exceeded the size limit.
Click the failed step and choose Retry to re-run just that step. The workflow picks up from the failure point, not from the beginning. In batch mode, you can retry all failed images at once.
Viewing per-step results
Click any completed step on the canvas to see what the image looked like after that step. This is great for debugging. If the final output looks wrong, step through the pipeline to find where things went off track.
For text-output steps like alt text generation or product descriptions, the step panel shows the generated text instead of an image preview.
Saving workflow output
If your workflow ends with an output step (Save to Library, Save to Sirv, Push to Shopify), results are saved automatically as the workflow runs. No extra action needed.
If your workflow does not include an output step, you can still download results manually from the results panel. You get the final image from the last processing step, plus any intermediate results you want.
Credit costs
Each AI processing step in the workflow costs credits independently. A three-step pipeline (remove background + smart crop + upscale) costs the sum of those three tools per image. Free steps like depth map or review gates do not consume credits.
Before running a batch, the workflow builder shows an estimated total credit cost based on your step count and image count. The actual cost may be slightly lower if some images fail early or get filtered out by an AI Router or Image Review step.